mongoose54

def train(net, train_iter, val_iter, batch_size, epochs, ctx):
    learning_rate = 0.01
    wd = 0.002
    log_interval=100
    if isinstance(ctx, mx.Context):
        ctx = [ctx]
    trainer = gluon.Trainer(net.collect_params(), 'adam',
                            {'learning_rate': learning_rate})
    loss = gluon.loss.SigmoidBinaryCrossEntropyLoss()

mongoose54

mongoose54

mongoose54

mongoose54

mongoose54

mongoose54

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {

mongoose54

____________________________________________________________________________________________________
convolution2d_134 (Convolution2D (None, 2, 64, 80)     66          convolution2d_133[0][0]          
____________________________________________________________________________________________________
permute_5 (Permute)              (None, 64, 80, 2)     0           convolution2d_134[0][0]          
____________________________________________________________________________________________________
reshape_7 (Reshape)              (None, 5120, 2)       0           permute_5[0][0]                  
____________________________________________________________________________________________________
activation_6 (Activation)        (None, 5120, 2)       0           reshape_7[0][0]  

mongoose54

_author__ = 'Fabian Isensee'

from collections import OrderedDict
from lasagne.layers import InputLayer, ConcatLayer, Pool2DLayer, ReshapeLayer, DimshuffleLayer, NonlinearityLayer, DropoutLayer, Upscale2DLayer, Upscale3DLayer, BatchNormLayer, batch_norm
from lasagne.layers.dnn import Conv3DDNNLayer as ConvLayer, Pool3DDNNLayer as PoolLayer
import lasagne
from lasagne.init import HeNormal

def build_UNet(n_input_channels=1, BATCH_SIZE=None, num_output_classes=2, pad='same', nonlinearity=lasagne.nonlinearities.elu, input_dim=(128, 128), depth=32, base_n_filters=32, kernel_size=3, do_dropout=False):
    net = OrderedDict()

mongoose54

#to_categorical() acquired from Keras implementation
#dice_coef() inspired by Marco Jocic's implementation: https://github.com/jocicmarko/ultrasound-nerve-segmentation/blob/master/train.py


def to_categorical(y, nb_classes=None):
    """Converts a class vector (integers) to binary class matrix.
    E.g. for use with categorical_crossentropy.
    # Arguments
        y: class vector to be converted into a matrix
            (integers from 0 to nb_classes).